Native vs. Processed Data
Understand the two types of data ingested and their respective costs.
What content is always free (Native Data)?
Native Data is structured content from your connected integrations: messages, tasks, text content, calendar items, and code-related text. Native Data syncs with zero tokens and is unlimited on all plans (subject to the Fair Use Policy).
Categories (see astell.space/en/integrations for the live list of connected systems):
- Messages: Slack channels, threads, and DMs; Gmail conversations
- Calendar: Google Calendar events, meeting details, and attendees
- Tasks: GitHub issues and pull requests; Linear tasks and projects
- Code & development: source code files, PR discussions, commit messages, code review comments, repository metadata
- Workspace text content: Notion pages and databases; DingTalk chats
(Microsoft 365, meaning Outlook mail/calendar, Teams, OneDrive, SharePoint, and Word/Excel/PowerPoint, plus Salesforce, HubSpot, Asana, Monday, Figma, and Zoom are on the integrations roadmap; see the canonical integrations page above.)
What kinds of content consume tokens (Processed Data)?
Processed Data is attachments and media that need processing beyond plain-text syncing. It consumes tokens when ingested. (Plain text content in Notion pages or Google Docs syncs free as Native Data; it's the file attachments and rich media that are processed.) Common examples:
- Documents: PDFs, Word, PowerPoint/Slides, Excel/Sheets, and file attachments embedded in pages or messages
- Images: screenshots, photos, diagrams, scanned documents
- Audio: voice notes, meeting audio, call recordings
- Video: meeting recordings, demos, training videos
- Web pages: ingested web content processed for indexing
- Meetings: recorded meetings that require transcription and processing
Why does some content consume tokens when text syncing is free?
Non-text processing requires extra work to process and index content: documents are priced by the amount of content processed; video processing includes transcription, diarization, and frame analysis; audio processing includes transcription and diarization; embedded images inside documents are processed separately.
How does unlimited syncing work?
Once you connect an integration like Slack, Gmail, or GitHub, Astell syncs your existing content and then keeps new content up to date in real time. All Native Data is indexed and fully searchable, and this syncing is completely unlimited, it never consumes tokens.
Is "unlimited" subject to the Fair Use Policy?
"Unlimited" applies to Native Data syncing from integrations, search, MCP connectors, and collaboration features. Chat (every model) is token-based and draws from your monthly pool. Fair use mainly applies when usage becomes automated, unusually high, or harmful to platform stability. Sapling may also be throttled earlier to protect free-tier reliability.